聚类分析和逐步回归法在车道流量预测中的综合应用(英文)

Application of cluster analysis and stepwise regression in predicting the traffic volume of lanes

针对目前国内外大中城市中普遍存在的无检测器信号交叉口车道交通流信息难于获取的情况,基于信号控制交叉口车道之间的相关性,综合应用聚类分析和逐步回归法预测单点无检测器信号控制交叉口车道流量.首先应用聚类分析将单点无检测器信号控制交叉口的车道与有检测器信号控制交叉口的车道交通流量进行聚类,然后在聚类分析结果的基础上随机选取车道交通流量样本运用逐步回归法预测单点无检测器信号控制交叉口的车道流量,此方法经过南京市的具体车道流量数据验证.此类问题的解决,可广泛应用于城市交通流诱导系统以及交通控制系统.

Because of the difficulty to obtain the traffic flow information of lanes at non-detector intersections in most metropolises of the world,based on the relationships between the lanes of signal-controlled intersections,cluster analysis and stepwise regression are integrated to predict the traffic volume of lanes at non-detector isolated controlled intersections.First cluster analysis is used to cluster the lanes of non-detector isolated signal-controlled intersections and the lanes of all signal-controlled intersections with detectors.Then, by the results of cluster analysis,the traffic volume samples are selected randomly and stepwise regression is used to predict the traffic volume of lanes at non-detector isolated signal-controlled intersections.The method is tested by the traffic volume data of lanes of the road network of Nanjing city.The problem of predicting the traffic volume of lanes at non-detector isolated signal-controlled intersections was resolved and can be widely used in urban traffic flow guidance and urban traffic control in cities without enough intersections equipped with detectors.